Profile Builder¶

coding=utf-8

Build model for a dataset by identifying type of column along with its respective parameters.

class dataprofiler.profilers.profile_builder.StructuredColProfiler(df_series=None, sample_size=None, min_sample_size=5000, sampling_ratio=0.2, min_true_samples=None, sample_ids=None, pool=None, options=None)¶

Bases: object

Instantiate the StructuredColProfiler class for a given column.

Parameters

df_series (pandas.core.series.Series) – Data to be profiled
sample_size (int) – Number of samples to use in generating profile
min_true_samples (int) – Minimum number of samples required for the profiler
sample_ids (list(list)) – Randomized list of sample indices
pool (multiprocessing.Pool) – pool utilized for multiprocessing
options (StructuredOptions Object) – Options for the structured profiler.

update_column_profilers(clean_sampled_df, pool)¶

Calculates type statistics and labels dataset

Parameters

clean_sampled_df (Pandas.Series) – sampled series with none types dropped
pool (multiprocessing.pool) – pool utilized for multiprocessing

diff(other_profile, options=None)¶

Finds the difference between 2 StructuredCols and returns the report

Parameters

other (StructuredColProfiler) – Structured col finding the difference with this one.
options (dict) – options to change results of the difference

Returns

difference of the structured column

Return type

dict

property profile¶

update_profile(df_series, sample_size=None, min_true_samples=None, sample_ids=None, pool=None)¶

Update the column profiler

Parameters

df_series (pandas.core.series.Series) – Data to be profiled
sample_size (int) – Number of samples to use in generating profile
min_true_samples (int) – Minimum number of samples required for the profiler
sample_ids (list(list)) – Randomized list of sample indices
pool (multiprocessing.Pool) – pool utilized for multiprocessing

static clean_data_and_get_base_stats(df_series, sample_size, null_values=None, min_true_samples=None, sample_ids=None)¶

Identify null characters and return them in a dictionary as well as remove any nulls in column.

Parameters

df_series (pandas.core.series.Series) – a given column
sample_size (int) – Number of samples to use in generating the profile
null_values (dict[str, re.FLAG]) – Dictionary mapping null values to regex flag where the key represents the null value to remove from the data and the flag represents the regex flag to apply
min_true_samples (int) – Minimum number of samples required for the profiler
sample_ids (list(list)) – Randomized list of sample indices

Returns

updated column with null removed and dictionary of null parameters

Return type

pd.Series, dict

class dataprofiler.profilers.profile_builder.BaseProfiler(data, samples_per_update=None, min_true_samples=0, options=None)¶

Bases: object

Instantiate the BaseProfiler class

Parameters

data (Data class object) – Data to be profiled
samples_per_update (int) – Number of samples to use in generating profile
min_true_samples (int) – Minimum number of samples required for the profiler
options (ProfilerOptions Object) – Options for the profiler.

Returns

Profiler

diff(other_profile, options=None)¶: Finds the difference of two profiles :param other: profile being added to this one. :type other: BaseProfiler :return: diff of the two profiles :rtype: dict

property profile¶

Returns the stored profiles for the given profiler.

Returns: None

report(report_options=None)¶

Returns the profile report based on all profiled data fed into the profiler. User can specify the output_formats: (pretty, compact, serializable, flat).

Pretty: floats are rounded to four decimal places, and lists are
shortened.

Compact: Similar to pretty, but removes detailed statistics such as
runtimes, label probabilities, index locations of null types, etc.

Serializable: Output is json serializable and not prettified Flat: Nested output is returned as a flattened dictionary

Variables: report_options – optional format changes to the report dict(output_format=<FORMAT>)
Returns: dictionary report
Return type: dict

update_profile(data, sample_size=None, min_true_samples=None)¶

Update the profile for data provided. User can specify the sample size to profile the data with. Additionally, the user can specify the minimum number of non-null samples to profile.

Parameters

data (Union[data_readers.base_data.BaseData, pandas.DataFrame, pandas.Series]) – data to be profiled
sample_size (int) – number of samples to profile from the data
min_true_samples – minimum number of non-null samples to profile

:type min_true_samples :return: None

save(filepath=None)¶

Save profiler to disk

Parameters: filepath (String) – Path of file to save to
Returns: None

classmethod load(filepath)¶

Load profiler from disk

Parameters: filepath (String) – Path of file to load from
Returns: Profiler being loaded, StructuredProfiler or UnstructuredProfiler
Return type: BaseProfiler

class dataprofiler.profilers.profile_builder.UnstructuredProfiler(data, samples_per_update=None, min_true_samples=0, options=None)¶

Bases: dataprofiler.profilers.profile_builder.BaseProfiler

Instantiate the UnstructuredProfiler class

Parameters

data (Data class object) – Data to be profiled
samples_per_update (int) – Number of samples to use in generating profile
min_true_samples (int) – Minimum number of samples required for the profiler
options (ProfilerOptions Object) – Options for the profiler.

Returns

UnstructuredProfiler

diff(other_profile, options=None)¶

Finds the difference between 2 unstuctured profiles and returns the report.

Parameters

other (UnstructuredProfiler) – profile finding the difference with this one.
options (dict) – options to impact the results of the diff

Returns

difference of the profiles

Return type

dict

report(report_options=None)¶

Returns the unstructured report based on all profiled data fed into the profiler. User can specify the output_formats: (pretty, compact, serializable, flat).

Pretty: floats are rounded to four decimal places, and lists are
shortened.

Compact: Similar to pretty, but removes detailed statistics such as
runtimes, label probabilities, index locations of null types, etc.

Serializable: Output is json serializable and not prettified Flat: Nested output is returned as a flattened dictionary

Variables: report_options – optional format changes to the report dict(output_format=<FORMAT>)
Returns: dictionary report
Return type: dict

save(filepath=None)¶

Save profiler to disk

Parameters: filepath (String) – Path of file to save to
Returns: None

classmethod load(filepath)¶

Load profiler from disk

Parameters: filepath (String) – Path of file to load from
Returns: Profiler being loaded, StructuredProfiler or UnstructuredProfiler
Return type: BaseProfiler

property profile¶

Returns the stored profiles for the given profiler.

Returns: None

update_profile(data, sample_size=None, min_true_samples=None)¶

Update the profile for data provided. User can specify the sample size to profile the data with. Additionally, the user can specify the minimum number of non-null samples to profile.

Parameters

data (Union[data_readers.base_data.BaseData, pandas.DataFrame, pandas.Series]) – data to be profiled
sample_size (int) – number of samples to profile from the data
min_true_samples – minimum number of non-null samples to profile

:type min_true_samples :return: None

class dataprofiler.profilers.profile_builder.StructuredProfiler(data, samples_per_update=None, min_true_samples=0, options=None)¶

Bases: dataprofiler.profilers.profile_builder.BaseProfiler

Instantiate the StructuredProfiler class

Parameters

data (Data class object) – Data to be profiled
samples_per_update (int) – Number of samples to use in generating profile
min_true_samples (int) – Minimum number of samples required for the profiler
options (ProfilerOptions Object) – Options for the profiler.

Returns

StructuredProfiler

diff(other_profile, options=None)¶

Finds the difference between 2 Profiles and returns the report

Parameters

other (StructuredProfiler) – profile finding the difference with this one
options (dict) – options to change results of the difference

Returns

difference of the profiles

Return type

dict

report(report_options=None)¶

Returns the profile report based on all profiled data fed into the profiler. User can specify the output_formats: (pretty, compact, serializable, flat).

Pretty: floats are rounded to four decimal places, and lists are
shortened.

Compact: Similar to pretty, but removes detailed statistics such as
runtimes, label probabilities, index locations of null types, etc.

Serializable: Output is json serializable and not prettified Flat: Nested output is returned as a flattened dictionary

Variables: report_options – optional format changes to the report dict(output_format=<FORMAT>)
Returns: dictionary report
Return type: dict

save(filepath=None)¶

Save profiler to disk

Parameters: filepath (String) – Path of file to save to
Returns: None

classmethod load(filepath)¶

Load profiler from disk

Parameters: filepath (String) – Path of file to load from
Returns: Profiler being loaded, StructuredProfiler or UnstructuredProfiler
Return type: BaseProfiler

property profile¶

Returns the stored profiles for the given profiler.

Returns: None

update_profile(data, sample_size=None, min_true_samples=None)¶

Update the profile for data provided. User can specify the sample size to profile the data with. Additionally, the user can specify the minimum number of non-null samples to profile.

Parameters

data (Union[data_readers.base_data.BaseData, pandas.DataFrame, pandas.Series]) – data to be profiled
sample_size (int) – number of samples to profile from the data
min_true_samples – minimum number of non-null samples to profile

:type min_true_samples :return: None

class dataprofiler.profilers.profile_builder.Profiler(data, samples_per_update=None, min_true_samples=0, options=None, profiler_type=None)¶

Bases: object

Factory class for instantiating Structured and Unstructured Profilers

Parameters

data (Data class object) – Data to be profiled, type allowed depends on the profiler_type
samples_per_update (int) – Number of samples to use to generate profile
min_true_samples (int) – Min number of samples required for the profiler
options (ProfilerOptions Object) – Options for the profiler.
profiler_type (str) – Type of Profiler (“structured”/”unstructured”)

Returns

BaseProfiler

classmethod load(filepath)¶

Load profiler from disk

Parameters: filepath (String) – Path of file to load from
Returns: Profiler being loaded, StructuredProfiler or UnstructuredProfiler
Return type: BaseProfiler